R Integration in Dell Statistica -...

R Integration in Dell Statistica

Integration options and features to leverage

specialized R functionality in Dell Statistica and

Dell Statistica Enterprise Server solutions

© 2016 Dell Inc. ALL RIGHTS RESERVED.

This guide contains proprietary information protected by copyright. The software described in this guide is furnished under a software license or nondisclosure agreement. This software may be used or copied only in accordance with the terms of the applicable agreement. No part of this guide may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording for any purpose other than the purchaser’s personal use without the written permission of Dell Inc.

The information in this document is provided in connection with Dell products. No license, express or implied, by estoppel or otherwise, to any intellectual property right is granted by this document or in connection with the sale of Dell products. EXCEPT AS SET FORTH IN THE TERMS AND CONDITIONS AS SPECIFIED IN THE LICENSE AGREEMENT FOR THIS PRODUCT, DELL ASSUMES NO LIABILITY WHATSOEVER AND DISCLAIMS ANY EXPRESS, IMPLIED OR STATUTORY WARRANTY RELATING TO ITS PRODUCTS INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. IN NO EVENT SHALL DELL BE LIABLE FOR ANY DIRECT, INDIRECT, CONSEQUENTIAL, PUNITIVE, SPECIAL OR INCIDENTAL DAMAGES (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF PROFITS, BUSINESS INTERRUPTION OR LOSS OF INFORMATION) ARISING OUT OF THE USE OR INABILITY TO USE THIS DOCUMENT, EVEN IF DELL HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Dell makes no representations or warranties with respect to the accuracy or completeness of the contents of this document and reserves the right to make changes to specifications and product descriptions at any time without notice. Dell does not make any commitment to update the information contained in this document.

If you have any questions regarding your potential use of this material, contact:

Dell Inc. Attn: LEGAL Dept 5 Polaris Way Aliso Viejo, CA 92656

Refer to our website (software.dell.com) for regional and international office information.

Trademarks Dell, the Dell logo, Dell Statistica, and Dell Statistica Enterprise Server are trademarks of Dell Inc. Other trademarks and trade names may be used in this document to refer to either the entities claiming the marks and names or their products. Dell disclaims any proprietary interest in the marks and names of others.

Legend

CAUTION: A CAUTION icon indicates potential damage to hardware or loss of data if instructions are not followed.

WARNING: A WARNING icon indicates a potential for property damage, personal injury, or death.

IMPORTANT, NOTE, TIP, MOBILE, or VIDEO: An information icon indicates supporting information.


Updated 2016

Dell Statistica Version 13.1

http://www.software.dell.com/


R Integration in Dell Statistica 13.1 3

Copyright © Dell, 2016

Contents

Contents

Introduction ................................................................................................ 4

Overview .................................................................................................... 6

Basic architecture and features of R support in Statistica ..................................... 11

COM interface to R environment .................................................................. 12

R integration supports macros (R.svb / R.r) ..................................................... 13

R scripts as native Statistica macros .............................................................. 13

Retrieving results ..................................................................................... 15

R language extensions: Passing data to R and retrieving results ............................. 17

ActiveDataSet ......................................................................................... 17

Definitions of Functions: ............................................................................ 18

RouteOutput(x,name,description) ................................................................ 19

Uses() Functions ...................................................................................... 21

Executing R scripts from within Statistica Visual Basic ......................................... 23

R scripts are Statistica macros ..................................................................... 23

R support in Statistica Enterprise Server .......................................................... 29

Creating R-based Statistica Workspace nodes .................................................... 33

Integrating R into Statistica Enterprise and Statistica Enterprise Server ................... 35

Creating R-based analysis configurations ........................................................ 35

Calling R scripts from SVB analysis configurations ............................................. 37

Summary ............................................................................................... 37

Final comments .......................................................................................... 38

Error handling ......................................................................................... 38

Strengths and limitations ........................................................................... 38

About Dell ................................................................................................. 39

Contacting Dell ............................................................................................ 39

Technical support resources ............................................................................. 39




1

Introduction

R is a programming language and environment for statistical computing (www.r-project.org). Most of the R

environment and its source code are currently available under the GNU GPL license (www.r-

project.org/licenses).

NOTE: None of the components of the R environment constitute unrestricted freeware. Instead, they

are available only under the terms of specific licenses. If you intend to download those applications,

you must accept prior to downloading and you must comply with their terms. Also, those licenses can

change over time. Thoroughly familiarize yourself with the terms every time you download any

components of the R environment.

Statistica can interface with third-party applications, such as Microsoft Office, through certain standards-

compliant channels.

EXAMPLE: COM is an interface for interprocess communication, which is built into Microsoft operating systems.

Even though R itself does not provide a COM interface, Statistica has created a separate connector application,

COMadaptR, licensed under LGPL (>=2.1) (with parts under GPL2) that facilitates communication to the R

world. This package is based on GPL2/LGPL2 version of an earlier application, the statconnDCOM library.

Statistica can interface with R via COMadaptR and, if this library is installed on the system, allows you to open

R scripts and submit them for execution to R from within the Statistica environment. It facilitating

bidirectional data transfer and presentation of resulting outputs through a user-adjustable macro that is

executed in Statistica in place of the R script.

This interface makes it possible for all Statistica products, ranging from Desktop to Enterprise and Web

solutions, to provide comprehensive support for interaction with the R platform on systems where necessary

third-party components and libraries are present, allowing you to:

run R scripts within the Statistica environment, sending results to Statistica reports, workbooks, and

graphs

process Statistica data sets in R and import tabular results from R into Statistica spreadsheets

call R from Statistica Visual Basic (SVB) to create new functionality that leverages R libraries

utilize R in Statistica Enterprise, Statistica Server, and Statistica Workspaces

NOTE: You are responsible for ensuring compliance with terms of all applicable licenses for R and all

components of the R environment. Always carefully review all the license agreements before accepting

them, as they can change over time. Those products are external to the Statistica environment, and are

not covered by any Statistica license agreements.

Automatic installation of the COMadaptR support library is included in Statistica Version 12.0 SP3 and above,

or is available with V12A maintenance update UPD008. When the system detects that an R installation is

present, it will offer you the opportunity to automatically install the COMadaptR package in the R

environment. This package is included in Statistica distribution.

You can verify that the COMadaptR library is installed correctly, and that all its dependencies are satisfied, by

running one of the R examples that accompany Statistica. Once all the required third-party components are

http://www.r-project.org/

http://www.r-project.org/licenses

http://www.r-project.org/licenses




installed, you should be able to open and execute R scripts within the Statistica environment. To locate the

examples, on the File tab, click Open Examples and browse to the R folder.

R is highly extensible. Users can submit libraries (packages), implementing a set of functions, usually for a

specific area of their expertise/research. The R community maintains several centralized repositories, which

make hundreds of such packages readily available to all users over the Internet. Many of these packages cater

specifically to highly specialized audiences with particular data analysis needs.




2

Overview

This document provides a detailed description of features that make the diversity and power of R fully

available to users of all Statistica solutions. Furthermore, these features enable you to combine the unique

capabilities of both Statistica and R platforms.

You can run Native R scripts directly within Statistica, Statistica Enterprise, and Statistica Enterprise

Server.

You can retrieve R output as native Statistica spreadsheets and graphs, and manage them via highly

flexible Statistica workbook containers.

Enterprises can now use the specialized routines and capabilities of R with Statistica, Statistica Enterprise, and

Statistica Enterprise Server for these functionalities:




Add new R-based modules

Leverage Statistica’s superior features to handle output from R

Graphics

Flexible spreadsheets

Convenient workbook containers for various document types




Integrate R into Statistica Enterprise

Make specialized R functionality available as reusable analysis templates to users not familiar with the

R language, in a secure, role-based enterprise analysis system.

Add R-based analytic nodes to Statistica Data Miner

Leverage all R capabilities inside Statistica Workspaces.




Build scalable R servers

Use Statistica Enterprise Server to handle security and load balancing

Use it to take advantage of multiple-processor servers to run R for demanding and/or validated

enterprise applications.




3

Basic architecture and features of R

support in Statistica

R support in Statistica creates an integrated Statistica-R platform that enables you to run R programs (scripts)

directly inside Statistica. By doing so, you can take full advantage of these specialized capabilities available in

R:

Enables users to run R

scripts as is, and

retrieve the results

into Statistica reports

All R console output is copied into the report, and R commands are highlighted.

Plots generated by the script are automatically embedded in the report as

scalable images.

These plots are also replicated as Statistica graphs (scalable metafile images are

placed inside these graphs), thus enabling annotation using powerful graphical

facilities in Statistica

The graphs can be printed or exported into a variety of image formats.

The reports can be edited, printed, and saved as PDF files.

Provides R language

extensions functions

for R scripts run from

the Statistica

environment

Transfer data from Statistica spreadsheets into R data frames.

Extract tabular data from R variables into Statistica spreadsheets. These results

spreadsheets (as well as all graphs produced by the script) are returned (routed)

according to Output Manager settings, into Statistica workbooks.

Executes R scripts as

native macros from

within Statistica

Visual Basic (SVB)

programs

Scripts can be parameterized with a Collection of objects (numbers, strings,

arrays, additional R code or overridden R functions, spreadsheets), which are

mapped to R variables accessible to the script. This approach provides fine-

grained control over scripts’ behavior in repeated runs or when they are used as

the backend for custom Statistica modules.

By default, all script output is routed by Statistica Output Manager. Alternately,

scripts can be executed using a method that returns its output as a Document

Collection, giving developers an easy way to extract specific analysis results that

could be used for further processing, such as input data for further analyses in

Statistica or in R, or for graphing.

Taken together, these enhancements not only enable you to run R scripts directly in the Statistica desktop

environment, but also provide a way to embed specialized R functionality into these features:

SVB programs

Custom interactive analysis modules

Workspace nodes

Enterprise analysis configurations

You can also offload such scripts to Statistica Enterprise Server for server-side processing.




Enterprise customers have the opportunity to integrate R into validated Statistica Enterprise solutions, such as

those used by FDA-regulated industries, or to provide a very powerful Statistica Enterprise Server-based R

server environment.

CAUTION: Developing and debugging R scripts: The R support in Statistica was not designed to supply

a complete R development and debugging environment. The console application and tools supplied with

standard R installation perform those functions very well and are already familiar to R users and

developers.

COM interface to R environment In order for R integration in Statistica to transfer data between Statistica and R, the R environment must be

installed in one of these locations:

On the same computer

On Statistica Enterprise Server

On the server for Enterprise solutions

The latest version of the R environment can be obtained from the CRAN website (http://cran.r-project.org).

To access the R environment, Statistica uses the COMadaptR library distributed under GNU Lesser Public

License. This library consists of two components:

One acts as an R COM Server.

The other is used for callbacks from R to Statistica.

Once R is installed, Statistica will automatically detect it. The next time you run Statistica, a dialog box will

be displayed.

1 Select to enable R Integration.

2 Click Yes, and Statistica will automatically install the COMadaptR library and register one of the

components with the Windows registry.

NOTE: Depending on your operating system version and settings, the system may ask you to confirm

these actions, possibly requesting administrator credentials.

3 Complete COMadaptR library installation by accepting the installer’s default settings.

You are now ready to use R within the Statistica environment. The COMadaptR library is independent from

Statistica and will remain on your computer until it is uninstalled manually.

If you do not need R Integration enabled, you can suppress the dialog box. The Integrating R into Statistica

dialog box will display when you start Statistica again.

1 If you do not wish it to display, select the Do not ask me again check box.

2 You can enable it again by selecting the corresponding check box on the Documents: Macros tab of the

Options dialog box.

The COMadaptR library provides a simple yet powerful COM (Component Object Model) interface to the R

environment. This interface can be used directly by SVB programs in Statistica.

An example of such a use case is included in the Statistica application, located in the Examples\R\Dose

Response folder. To use it, open Direct Interface To R via COM.stw and run the embedded SVB macro.

http://cran.r-project.org/




CAUTION: Usage of such an interface directly by end users is very ineffective, sometimes unproductive,

and usually inflexible. It may also significantly degrade the overall performance of interactions with R if

performed incorrectly.

Therefore, the following architectural extensions have been added to the Statistica platform to provide a

seamless and effective R Integration experience for end users. The example mentioned above also

demonstrates the significant reduction in the efforts required to implement the same analysis using the new

built-in features. In Statistica, simply open and run DoseResponse.r.

R integration supports macros (R.svb / R.r) Statistica installation includes two element. R.svb and R.r contain the support code required to manage

interactions through COM in Statistica and the R side respectively.

R script called R.r Support script R.r implements Statistica-specific R language extensions

functions.

One of its primary responsibilities is translating Statistica spreadsheets to

R data frames and back.

When an R script is executed in Statistica, the support macro parses the script and executes these processes:

1 Transfers data and script parameters

2 Submits script content to the R environment

3 Manages error conditions

4 Handles script outputs, ensuring that they are properly transferred back to Statistica.

CAUTION: Although the support code in R.r is write-protected by default, you can inspect it and

modified or enhanced it to support new functionality required for specific use cases, although you

do so at your own risk.

Statistica Visual Basic macro

called R.svb

The R.svb macro supports standalone execution to simplify debugging and

testing of modifications.

R scripts as native Statistica macros Statistica recognizes .R (and .S) file extensions as R scripts. Open these files by selecting the File tab and

clicking Open.

It also registers (at the operating system level) .R/.S files as Statistica macros, so you can open these files in

Statistica from a file browser by double-clicking on them.




R scripts are displayed in slightly modified Statistica Visual Basic Macro windows. These windows actually

contain two scripts, the R script itself and the R Integration Support Macro (R.svb), accessible through two

tabs in the upper-left corner (circled in the next image).

Limited R code highlighting is available (strings, language extensions, VB-style comments).

Creating a new R script

You can also create a new R script within Statistica.

1 Select the File tab.

2 Select New.

3 Under the Classic New Document Dialog heading, click New Document to display the Create New

Document dialog box.

4 Select the Macro tab.

5 Select the R (requires R Statistical Environment) option button.

6

NOTE: This option is available only if R integration support is installed on your machine.




7 Click OK to open an empty R script window.

8 You can now type or paste in an R program.

R Integration support also includes an optional text file called R.inc, located in the default installation

directory along with R.svb. The contents of the file are copied to the beginning of each new R script created

in this manner.

Running R scripts within Statistica

To execute the R.svb macro for the currently active R script, on the Macro tab, in the Debug group, click the

button or press the F5 key on your keyboard.

Although breakpoints are not supported for the R script itself, you can set breakpoints and debug the R.svb

macro on the second tab of the modified Statistica Visual Basic Macro window while running the R script.

NOTE: In order to take advantage of R Integration features described in this document, R scripts should

be executed from within Statistica.

CAUTION: Although you can develop and debug complex R programs within this environment, it was not

specifically designed for these purposes. The R environment itself is better suited for such activities.

Retrieving results The minimal output produced during execution of an R script is a Statistica workbook that represents an R

console session. It will include highlighted commands and any output generated by the R environment. This

kind of report is produced even if the script is empty. The contents of this report can be edited and

manipulated in the same way that you would edit any other Statistica report.

What happens to all plots created during an R session:

They are automatically transferred into the Statistica environment as Windows metafiles (vector

graphics format). They can be resized without loss of quality.

They are placed into the R console session report. They create a natural flat report of the R session,

with embedded plots tied to the graphics commands, which is not easily achieved in the R environment

itself.

They are replicated as Statistica graphs that become a part of the R script output.




The metafile images are embedded into graph objects as locked background.

You can annotate R plots in Statistica, using a familiar point-and-click interface, with a set of text and drawing

objects such as lines and arrows, rectangles and ellipses, polygons and pattern/color fill areas, etc.

Since these annotations are anchored to relative positions in the plot area, they will remain correctly attached

to the plot if the graph is resized.

You can flexibly design and further enhance these graphs using Statistica graphics tools. You can save them in

other formats (JPG or GIF), or print them to PDF files.

Since the individual R plot components (the structural elements of the plot) are not accessible for

manipulation in Statistica graphs, the rich capabilities of Statistica for creating and then further editing graphs

(scaling, point markers, fit lines, etc.) are not available.

However, integration between R and Statistica lets you extract data from R. You can then render important

graphs inside the Statistica environment by writing Statistica Visual Basic macros that will execute R scripts,

extract results, and then post-process those results as necessary, which will be illustrated later.

Unique Statistica capabilities

R script output consists entirely of native Statistica objects. These objects are properly routed by Statistica

according to Output Manager settings. Depending on user selection, they could be placed into standalone

windows or in a workbook.

These are examples of ways you can manage each of these objects, using the extensive capabilities of the

Statistica platform:

Annotate

Store in a (compressed) workbook

Export as Microsoft Office documents

Print

Save as a PDF file

Convert to one of many popular formats

Archive as version-controlled, auditable items in the Statistica Document Management System

Share among other users in a web-based, client-server Statistica Enterprise Server environment




4

R language extensions: Passing data to

R and retrieving results

The R Integration Support Macro (R.svb) implements several extensions to R language. These keywords and

functions can be used inside R scripts executed within the Statistica environment. These extensions enable

scripts to pass data to R and retrieve results from the R environment.

IMPORTANT: The R language is case sensitive; therefore, R language extensions for Statistica are also

case sensitive. Extensions are only recognized by the Statistica environment when typed exactly as

shown.

Tabular data represented in Statistica in the form of spreadsheets are mapped into the equivalent R

structures: data frames.

Mapping preserves as much information as possible for both formats, as illustrated by these examples:

Text label variables in a

spreadsheet

Factor objects in a data frame

Missing data values NA indicators

Data types Variable and case names

ActiveDataSet The ActiveDataSet keyword has been adopted from the Statistica Visual Basic language. It performs the same

function in R scripts by referencing the active Statistica data spreadsheet.

In the desktop Statistica environment, the active data set usually means the top-most visible spreadsheet that

can act as a data source. However, it can also be a spreadsheet in a workbook selected as Active Input.

This concept is redefined and extended for server-based environments (Statistica Enterprise Server,

Enterprise), but the keyword is still valid and refers to the corresponding server-side mapping of the active

data source.

If no active data set is defined/available, the R script that uses it will fail. The same is true for SVB macros.

Become

Mapped To

Transferred both ways




If the R Integration Support Macro encounters the ActiveDataSet keyword in the R script, it transfers the actual

Statistica active data set into the R environment and assigns it to a variable of the same name. Therefore, this

keyword represents a data frame variable and can be handled as such in the script.

EXAMPLE:

Definitions of Functions: This section explains the following functions: Spreadsheet(path.or.object, [applySelectionConditions],

[recodeTextLabelAsFactors], [recodeMissingDataToNA], [getCaseNames], [getTextValues], [attachObject],

[username], [password], [connectionstring], [stationname])

Function Values/Description Examples and Notes

Spreadsheet() extension Used to load a specific Statistica data

file into R and to transfer the data in

that file to an R data frame

ActiveDataSet Keyword Similar to Spreadsheet() extension

path.or.object Character vector

(literal or variable) or a

valid COM object

pointing to a Statistica

spreadsheet

C:/data/sample.sta

Enterprise://SomeFolder1/Som

eFile.sta

Adstudy

[applySelectionConditions] TRUE (default) or

FALSE flag Indicates whether selection conditions

defined on spreadsheet should be

employed

[recodeTextLabelCodesAsFactors] TRUE (default) or

FALSE flag Indicates whether text labels defined

for variables should be treated as

factors or text

[getCaseNames] TRUE (default) or

FALSE flag Indicates whether row names should be

fetched

[attachObject] TRUE or FALSE

(default) flag Indicates whether spreadsheet should

be attached to data frame as an Object

attribute

[username] character strings

(default value: “”)

Used to fetch spreadsheets from the

enterprise system when user does not

have an integrated login [password]

[connectionstring]

Spreadsheet()

Like the ActiveDataSet keyword, the return value of the Spreadsheet() function should be treated as a data

frame variable with the contents closely matching that of the corresponding Statistica spreadsheet.

ActiveDataSet[1:5] # display a subset of active dataset (variables 1 through 5)

str(ActiveDataSet) # view the structure of the data frame

plot(ActiveDataSet$MEASURE01) # plot "MEASURE01" variable values




One useful feature supported by this function is the use of default search paths for spreadsheet files that are

specified only as simple file names. This means that if the function parameter consists only of a file name,

such as Spreadsheet(“some.sta”), R integration support code looks for this file in several locations:

Support code will also append the default .sta file extension if one is not present. Therefore the following

options are available:

R scripts can reference the accompanying data sets (placed in the same folder) simply by name.

Spreadsheets that are included in every Statistica installation as demonstration (Example: data sets)

can be referenced by name in much the same way as built-in R data sets.

EXAMPLE:

Spreadsheet("c:\myfiles\mydata.sta") # display the dataset in R console

Spreadsheet("\\server\share\data.sta") # same, but read from a network share

Spreadsheet("thisdemo.sta") # file in the same folder as this script

Spreadsheet("enterprise://folder/thisdemo.sta") # file in STATISTICA Enterprise

# plot a histogram for MEASURE05 variable from STATISTICA example dataset Adstudy:

hist(Spreadsheet("Adstudy")$MEASURE05)

...

advert = Spreadsheet("Adstudy")$ADVERT # retrieve a data frame variable (factor)

is.factor(advert) # => [1] TRUE

levels(advert) # => [1] “PEPSI” “COKE”

RouteOutput(x,name,description) RouteOutput transfers various types of data from the R environment into Statistica spreadsheets:

x a data frame, matrix, array, number, or a string

name character vector containing name (in Statistica) for the

object being routed

description character vector containing a description for the object

being routed

SpreadsheetName, SpreadsheetHeader literal string, such as “A Frequency Table”

Although the function was introduced to retrieve tabular data (such as data frames, matrices, or arrays) into

spreadsheets, single-value data, such as numbers or strings, can be passed as well. They will be placed in

single-cell spreadsheets. x can be an R variable or a literal value.

The extension, RouteOutput(), is similar to the equivalent Statistica Visual Basic function. The results

spreadsheets recreated by this function in the Statistica environment become the standard output of the R

script/analysis. They follow Output Manager settings in Statistica.

First, it checks the folder where the R script itself is located (if it

was saved to disk).

Next, it checks the Examples\Datasets folder for the

current Statistica installation.




NOTE: In Statistica:

Select the Tools tab.

Click Options.

In the Options dialog box, select the Analyses/Graphs: Output Manager tab.

The results are routed either to individual windows or to a workbook (or multiple workbooks) for each analysis,

with optional output reports, such as a Microsoft Word document. The most popular setting is a single results

workbook.

Optional parameter name and description specify the name and header of the resulting spreadsheet. Provide a

value for the spreadsheet name for visual distinction in the tree view of the results workbook.

NOTE: R plots transferred into Statistica as native graphs do not require explicit output routing. All

plots generated during a script run are automatically transferred and routed according to Output

Manager settings.

Many functions in R, specifically the ones that perform statistical modeling, represent their results as

structured objects, sometimes of significant complexity.

These objects cannot be reduced to a single table, and therefore cannot be handled by the RouteOutput()

extension. They could be automatically searched for tabular components, but since the object structures are

specific to a particular method, such an approach would generate a significant amount of “junk” output.

However, since the results (the actual data of interest) are either stored in such objects as tabular

components or produced by applying an object’s method to some input data, this limitation does not pose any

problems. Particular results can be easily extracted from such a statistical model object and routed back to

Statistica.

EXAMPLE:

# build a statistical model (Dose Response Curves)

model <- multdrc(SLOPE ~ DOSE, CURVE, data = PestSci)

...

str(model) # display the complex structure of this model/results in R console

...

# transfer results to Statistica

RouteOutput(coefficients(summary(model)), "Summary", "Parameter Estimates")

RouteOutput(residuals(model), name = "Residuals", header = "Some Header")

RouteOutput(anova(model), "ANOVA") # inherits the default heading from data frame

RouteOutput(logLik(model), "Log-Likelihood")

RouteOutput(ED(model, c(10, 50, 90)), "Relative Effective Doses (10%/50%/90%)")




The results for this example may look like this:

Uses() Functions This section covers the following functions: Uses(package,[lib], [repos], [quiet], [attachImports]).

The Uses() function ensures that the respective package (library), named package, and its dependencies

(packages that are required in order for this package to run) are:

1 installed on the computer/server where the R script is running

2 loaded into the R environment automatically

If some of these libraries are not present, they will be installed from CRAN repository (http://cran.r-

project.org) and then loaded.

Functions Descriptions

package Character vector containing name of package, such as drc

lib Character vector containing path to library directories to install package if

needed

repos List of repositories to use to download package if needed

quiet TRUE or FALSE (default) flag indicating whether interactive dialog boxes should

be suppressed during package installation

attachImports TRUE OR FALSE (default) flag indicating whether to attach imports from the

package and its dependencies to the current R session

R defines several levels for specification of package dependencies.

One of them, Imports, lists packages with namespaces that have to be loaded for the package to run.

However, the namespaces do not need to be attached to the current/global environment, so the package

accesses such imported packages using namespace-qualified names, such as tseriesChaos::mutual(x) instead

of mutual(x).

Three ways to use some of these namespace-enclosed functions in your own R code:

Explicitly use namespace qualification [tseriesChaos::mutual(x)].

Attach the library to current environment [Uses(“tseriesChaos”) or library(tseriesChaos)].

Set the optional Uses() parameter attachImports to TRUE , which will attach all the imports of the

package.






NOTE: This extension is not necessary for interaction between Statistica and R. You can implement it

within the R language itself. It simplifies the process of conditional library installation and loading by

encapsulating it in a single call.

EXAMPLE:

Uses("drc") # make sure that the respective package is installed and loaded

...

DR <- multdrc(SLOPE ~ DOSE, CURVE, data = PestSci) # call the package methods

This program fits dose response curves to the respective variables of the built-in PestSci data set by calling the

multdrc function defined in the “drc” package. Uses(“drc”) ensures that the function is available by installing

and loading the package, if necessary.




5

Executing R scripts from within

Statistica Visual Basic

Leverage specialized R functionality within Statistica by calling R scripts from inside a Statistica Visual Basic

(SVB) macro to accomplish the following tasks:

Build new modules using the SVB User Interface library and methods implemented in R scripts.

Create Statistica Enterprise SVB analysis configurations.

Create Statistica Workspace nodes (for Statistica or Statistica Enterprise Server) that leverage R.

To provide any non-trivial functionality within an R script in such use cases, you must parameterize that script

with user-selected parameters, variables lists, input spreadsheets, etc. Statistica provides a simple and

powerful way to pass such parameters to R scripts.

R scripts are Statistica macros In almost all cases, Statistica treats R scripts in the same way as it treats native SVB macros. This rule applies

to the Statistica Object Model as well. Now Macro objects in SVB programs can represent R scripts. Therefore,

R scripts can be created, opened, edited, saved, and executed from within SVB scripts.

R functionality is available in Statistica Enterprise analysis configurations and Statistica Workspace nodes,

since they are SVB-based.

Existing R script files can be opened with Macros.Open(“path\to\some.r”) or created on-the-fly with

Macros.New() and Macro.Code.

In the latter case, Statistica needs help in distinguishing R scripts from SVB macros, which can be achieved by

either of two methods:

Specify the name for a new macro with the .R extension, even if you are not going to save it on disk.

Explicitly set Macro.Scripting to 5 (R Macro Type).

Run the scripts by calling Macro.Execute.

Create

Open

Edit

Save

Execute




NOTE: The Macro.Scripting type for R scripts is 5. It will later be mapped to a symbolic constant.

EXAMPLE:

Sub Main

Dim R As New Macro

R.Code = "ActiveDataSet" ' simple R script created on-the-fly

R.Scripting = 5 ' R Macro Type = 5

R.Execute

End Sub

This Statistica Visual Basic macro runs a simple R script containing only a single command ActiveDataSet. As

described in the previous section, this script is an R language extension for Statistica that will transfer (and in

this case display) the currently active Statistica data file in R.

EXAMPLE: If you run this macro after opening the example data file Exp.sta, a listing of that file will be

displayed in a report window that represents the R console session, as shown below:

Passing parameters to R scripts: Collection Object

A Collection is a COM library, introduced with Statistica version 8 MR3, that implements generic Collection

objects. Such objects enable users to store keyword-tagged Variant values. Variants are COM structures that can

store one of many supported data types, including arrays and references, to other COM objects, such as

spreadsheets.

So in R integration, you can now store {parameter name = parameter value} pairs in such a Collection, then

pass this parameter set to Statistica prior to R script execution, to achieve script parameterization. The R

Integration Support macro will transfer all the parameters from the Collection to the R environment as R

variables.

The variables are named using the respective parameter keyword tags, which makes it possible for the script that

is executed in this environment to reference all these variables (script parameters) directly by name.

R integration supports the following types of named parameters:

Strings

Numbers

Arrays

Spreadsheets

Support for Statistica spreadsheets is particularly useful.

EXAMPLE: Such a parameter can override the ActiveDataSet keyword, allowing a single script to serve as an R

engine for an interactive module, Enterprise analysis configuration, and Workspace node implementations.




NOTE: String parameters without a keyword tag are treated as R code that should be executed prior to

execution of the script itself. This is analogous to SVB hidden code and can be used to, for example,

define a common set of new functions or global variables/constants.

In order to use Collection objects, SVB scripts must include a reference to the Statistica Collection Library.

Add it via Tools > References while editing the macro.

The Collection object has several generic properties and methods that are needed to manipulate the contents

of a Collection:

Count

Exists (key)

Item (key), which adds a member to the collection with the specified key or returns an existing

member of the collection with the specified index or key

Add (item, [key]), which adds a member to the collection with an optional associated key

Remove (Index or key)




Remove All

However, due to the use of so-called default object properties (Item is the default property of a Collection

and it returns its Value property by default), interaction with a Collection object is reduced to intuitively

clear assignment operations:

EXAMPLE:

Dim param As New Collection

param("number") = 57

param("string") = "A string sample..."

Once you have assembled the parameter Collection, execute the parameterized R script by calling

Macro.ExecuteWithArgument(Parameters As Collection).

EXAMPLE:

Dim s1 As New Spreadsheet, s2 As New Spreadsheet ' ... populate s1, s2 with data

var1 = Array("CASE 1", "CASE 2")

var2 = Array(1, 2, 3, 4, 5)

' * don't use spaces in parameter names

' * some names are "locked" and can't be used [e.g. 'text', 'str', 'sample']

Dim param As New Collection

param("number") = 57

param("string") = "A STRING sample..."

param("string_array") = var1 ' add items with an assignment operator

param.Add(var2, "number_array") ' OR using explicit Add() method

param("SomeSpreadsheet") = s1

param("ActiveDataSet") = s2 ' override the value of 'ActiveDataSet' keyword

' string parameters without associated keys will be treated as R code and

' will be executed before the script - an analog of SVB 'hidden code'

' * define a function that will be available to the R script

param.Add("func <- function(x) { cat('Called func(x) with x =', x) }")

' * another way to define a global constant or variable

param.Add("STATISTICA.Version = '" & Version & "'")

' now run the R script with this collection of parameters

' (parameters become R variables – the script can reference them by name)

Dim m As Macro

Set m = Macros.Open(MacroDir & "\parameterized.r")

m.ExecuteWithArgument(param)




Managing script results Both Macro.Execute() and Macro.ExecuteWithArgument() methods handle R script output (spreadsheets,

graphs, reports) in exactly the same way as the output from SVB macros. The output is routed by Statistica

according to Output Manager settings.

EXAMPLE: The output is placed into a results workbook.

This approach might be sufficient for many use cases, but is not flexible enough in situations where the output

from an R script is treated as an intermediate result for subsequent processing in Statistica.

EXAMPLE: An SVB script implementing a custom interactive module might implement a multiple stage analysis

workflow, calling into R, analyzing the results, requesting additional inputs from the user, and calling into R

repeatedly with previous results as inputs. Or, an SVB macro could use tabular results returned from R to

produce graphs that are more expressive and flexible than what R is able to generate.

Since the items processed by the Statistica Output Manager are intended to be a final representation of

analysis results, accessing them individually for further processing is not a trivial task. You would need to know

the type of representation selected (single or multiple workbooks, individual windows) and programmatically

iterate through the respective set of output items (windows, workbook contents represented as a tree),

searching the item of interest by name.

Macro.ExecuteNoRouteOutput([Parameters As Collection]) As StaDocCollection

For cases such as this, the above method has been added, where R script output is not routed at all. As a

result, the script runs silently, and all of the output from the script is collected into a StaDocCollection

object. This standard Statistica container for documents such as spreadsheets, graphs, reports, or workbooks is

returned as the result of script execution. It also accepts an optional Collection for parameterization.

SVB macro developers can easily access individual components of R script output. For instance, you can

extract individual cell data from spreadsheets or create a complex graph based on multiple columns from

several spreadsheets.




EXAMPLE:

Dim m As Macro

Set m = Macros.Open(MacroDir & "\some.r")

Dim Routput As StaDocCollection

Set Routput = m.ExecuteNoRouteOutput() ' silent, no output displayed

' do something with "Routput": extract and operate on individual documents

' trivial case: results are routed (displayed), as if from Macro.Execute()

RouteOutput(Routput, "R Script Results").Visible =True

More examples

We have demonstrated all the functional components required to build custom applications within the

Statistica platform that can take advantage of the specialized functionality available in R.

All installations of Statistica contain a set of examples that provide a detailed demonstration of the R

integration features. You will find these examples in the [Statistica]\Examples\R folder. These examples can

also be used as templates for your own development.




6

R support in Statistica Enterprise Server

Statistica and Statistica Enterprise Server are based on identical Statistica libraries, and support identical

functionality. This is true for R support in Statistica as well. You can execute R in Statistica Enterprise Server

in much the same way as from desktop (thick-client) Statistica.

EXAMPLE:

1 Log on to Statistica Enterprise Server.

2 Select Tools > Submit/Execute a Custom Script.

3 Type in the macro shown below (available as standalone.r in the Examples\R folder).

NOTE: Save the script with the .R extension.

4 To run the script, click Save and Execute or use the File menu to open and run it.




NOTE: This script does not require a data file. If you are using ActiveDataSet or Spreadsheet()

extensions, keep in mind that the data sources must be located on the server:

Spreadsheet() should reference data sets by Statistica Enterprise Server URL paths.

ActiveDataSet will be mapped to the spreadsheet that you open in Statistica Enterprise Server.

This rule can be overridden by a parameter of the same name passed to the R script from SVB. Also, in

order to execute any scripts in Statistica Enterprise Server, you must have the proper permissions on

the server.

Statistica Enterprise Server is a powerful R server

Statistica Enterprise Server was designed as a powerful and flexible server/web-based analytical platform,

relying on the Statistica Visual Basic engine for the diversity of its functionality, as well as extensibility.

R scripts are handled by Statistica Enterprise Server in much the same way as standard SVB macros. The third-

party components that the Statistica platform relies upon to provide its R runtime environment (such as R

library and R COM Server library) are also well suited for handling multiple simultaneous R sessions.

Thus, Statistica Enterprise Server represents an ideal platform for a powerful and flexible multiprocessor R

server that can handle a large number of users, providing security, scheduling, load balancing, etc.

Offloading R scripts from Statistica to Statistica Enterprise

Server

Like SVB scripts or Workspaces, R scripts can be offloaded from a desktop environment for processing on

Statistica Enterprise Server. This capability is particularly useful when an R script is expected to require

significant computing resources. You can submit such scripts to the server and retrieve the results at a later

time.

EXAMPLE: Allow them to process overnight.

If your installation is set up to work with Statistica Enterprise Server (Tools > Options: Server/Web), you can

offload any R task by selecting Server > Offload Task….




Options are also provided to transfer the data file to the server side. These options are described in detail in

the Statistica documentation.

After clicking OK, the respective R script will execute on Statistica Enterprise Server. It might be scheduled to

be executed rather than be executed immediately, depending on the server load.

The progress of the analysis can be monitored via the Server > Task Status dialog box.




You can retrieve the results from the server via the Results button or you can double-click on the task. The

representation of the results of an offloaded task is equivalent to the same task running locally.




7

Creating R-based Statistica

Workspace nodes

You can use the techniques described in previous sections to implement a custom R-based Workspace node:

1 Edit the SVB node’s source to generate a Collection based on the node’s input parameters.

2 Use it to execute one or more parameterized R scripts.

3 Retrieve the results as a StaDocCollection, which becomes the output of the node.

EXAMPLE:

1 Open Examples\R\NonLinear Time Series\As Data Miner Node.sdm.

2 In the Workspace, right-click the analysis node.

3 Select Edit Code to review the underlying SVB code (NonLinearTimeSeries.svx).

NOTE: The corresponding .dmi file describes the node’s input parameters. The R script file

(NonLinearTimeSeries.r) performs nonlinear time series analysis using the tsDyn R package. This

script can serve as a template for your own R-based Data Miner nodes.

4 After running the node, double-click the Reporting Documents node to display the workbook.




This example demonstrates that the ability to incorporate specialized R functionality into Statistica Data Miner

provides very powerful opportunities for creating analytic workflows or Collections of Data Miner nodes and

node-browser configurations dedicated to highly specialized analyses (such as nonlinear time series, dose

response curve fitting, and so on).




8

Integrating R into Statistica Enterprise

and Statistica Enterprise Server

R Integration in Statistica would be incomplete without support for Statistica Enterprise and Statistica

Enterprise Server solutions. Now, R programs can be used as analysis templates, and R analyses can be

distributed to end users who are not familiar with VB or R programming.

The Enterprise platform provides numerous options and features to specifically enable the application of

templated analyses for validated applications, such as for manufacturers who need to comply with the

requirements of FDA 21 CFR Part 11, such as version control and audit trails. These features make R available

for deployment in mission-critical manufacturing applications.

Creating R-based analysis configurations The Statistica Enterprise system implements standard control charting and SPC analyses out-of-the-box. Non-

trivial analysis templates are usually created using Statistica Visual Basic, for example, by using Statistica

macro-recording capabilities. In addition, Statistica Enterprise and Statistica Enterprise Server now recognize

R scripts and can store and process them in the same way as SVB scripts.

When creating a new Analysis Configuration in Statistica Enterprise:

1 Select the R Analysis option.

2 Type, paste in, or load the respective R script.

3 All other options are identical to those available for SVB-based analysis templates.




Depending on the configuration of the Statistica Enterprise system, this reusable analysis template is now

available to the respective users of both Statistica and Statistica Enterprise Server environments. The results

of this template can be combined into standard reports.




Moreover, if the respective Statistica Enterprise installation is integrated with the Statistica Document

Management System, these R scripts can be locked down. For instance, you could version them with complete

audit trails to support FDA 21 CFR Part 11 requirements.

Calling R scripts from SVB analysis configurations R functionality can also be utilized from inside SVB analysis templates. They can retrieve the results from R for

further processing or display. In this case, R scripts that are loaded for the respective templated analyses in

Statistica Enterprise can be called in either of two ways:

They can be placed into a secure folder repository, from which they can be loaded, and where they can

be managed via Statistica Document Management System integrated with Statistica Enterprise.

They can be embedded into the SVB code. To use this method, assign R script text to the SVB macro’s

Code property.

Summary The following capabilities, made possible by the full integration of the R environment into Statistica

Enterprise, make Statistica Enterprise arguably the most powerful enterprise analysis system to date:

Utilize extensive and highly specialized functionality of R.

Leverage it inside a secure, enterprise-wide, role-based analysis system that insulates end users from

implementation details and that can be deployed into validated manufacturing environments.

Run R-based templates on the Statistica Enterprise Server platform.

Extract results into pre-formatted standard reports.




9

Final comments

The features provided in Statistica for integration with R are quite flexible. Using them, thousands of highly

specialized R functions and features become available to all Statistica solutions. However, if you plan to

exploit these features, please consider the following system limitations of the R platform.

Error handling Most of the error conditions generated within the R environment, such as syntax and runtime errors caused by

an R script or by the integration support libraries, such as broken R installation or missing components, are

intercepted and handled by Statistica. Developers can use error handling facilities available in either

environment.

EXAMPLE: On Error handlers in SVB macros calling R scripts.

However, R programs can occasionally crash or hang. In the latter case, program control does not return to

Statistica. Therefore, careful validation of the respective R scripts is crucial for enterprise-level deployment of

R analysis templates.

Strengths and limitations

A word of caution regarding the quality of R algorithms

R comes without warranty or guarantees. In practice, many (most) of the algorithms available in R are the

result of diligent work over many years by one or a few experts in their respective methodology or domain.

However, this does not mean that the software was created following rigorous software development lifecycle

methodology, or stringent standard operating procedures for software requirements gathering, design,

implementation, and testing.

Therefore, in order to build a mission-critical or validated application around a component that depends on R,

it is absolutely critical that you carefully validate all results for the use cases to which the software is to be

applied.

A word of caution regarding scalability, large data sets, etc.

Another caveat regarding R that needs to be considered before building solutions around R concerns its basic

architecture. Unlike Statistica, data in R must be, in practically all cases, resident in the computer’s memory.

This restriction, in combination with hardware-level and operating system-level memory limitations, may or

may not pose an obstacle for any one individual user, but will need to be considered carefully when building R-

based server applications accessible to multiple users.

A word of caution regarding consistency

Results generated using R might vary across versions of R, platforms (Windows, MacOS, Linux), and platform

bitness (32-bit vs 64-bit) for identical input values and identical programs. Statistica faithfully transmits to and

from R, and is not responsible for tracking, reporting, or adjusting for such changes.




About Dell

Dell listens to customers and delivers worldwide innovative technology, business solutions and services they

trust and value. For more information, visit www.software.dell.com.

Contacting Dell Technical Support:

Online Support

Product Questions and Sales:

(800) 306-9329

Email:

[email protected]

Technical support resources Technical support is available to customers who have purchased Dell software with a valid maintenance

contract and to customers who have trial versions. To access the Support Portal, go to

http://software.dell.com/support/.

The Support Portal provides self-help tools you can use to solve problems quickly and independently, 24 hours

a day, 365 days a year. In addition, the portal provides direct access to product support engineers through an

online Service Request system.

The site enables you to:

Create, update, and manage Service Requests (cases)

View Knowledge Base articles

Obtain product notifications

Download software. For trial software, go to Trial Downloads.

View how-to videos

Engage in community discussions

Chat with a support engineer

http://www.software.dell.com/

http://software.dell.com/support/

mailto:[email protected]

http://software.dell.com/support/

http://software.dell.com/trials

R Integration in Dell Statistica -...

Documents

Transcript of R Integration in Dell Statistica -...